Recognition Systems for Practical Applications

• Published in: Markov Models for Handwriting Recognition pp. 47 - 66
• Main Authors: Plötz, Thomas, Fink, Gernot A.
• Format: Book Chapter
• Language: English
• Published: London Springer London 07.09.2011
• Series: SpringerBriefs in Computer Science
• Subjects: Datasets; Evaluation; Frameworks; Organizations; Research groups; Toolkits
• ISBN: 9781447121879; 1447121872
• ISSN: 2191-5768; 2191-5776
• DOI: 10.1007/978-1-4471-2188-6_5

Aiming at a comprehensive overview of Markov-model based handwriting recognition this chapter focusses on the description of recognition systems for practical applications. After the theoretical aspects and key developments in the field have been surveyed, integration aspects and concrete evaluations of recognition capabilities are discussed. The chapter starts with a description of the most relevant data-sets. As usual for all experimental science, handwriting recognition research relies on the availability of high-quality sample data for training and evaluation purposes. According to the general shift of research efforts from online to offline handwriting recognition, the majority of systems described in the current literature is dedicated to offline recognition. Reviewing the literature, we identified seven major recognition systems. We concentrated on those systems that are still being maintained and further developed. In this chapter their key features will be described and performance figures will be given.